Showing posts with label genetic analysis. Show all posts
Showing posts with label genetic analysis. Show all posts

Thursday, February 1, 2018

Inexpensive screening for germline mutations to personalize treatment

Pritchard et al. last year discovered that certain rare germline mutations that interfere with DNA-repair mechanisms occur with greater than expected frequency in men with aggressive prostate cancer. A "germline mutation" means that it is inherited from one's parents and is part of a man's normal genetic profile, for better or worse. By contrast, a "somatic mutation" means that it occurs only in tumor tissue and not necessarily in normal cells. There are several genes responsible for repair of our DNA. Their job is to fix random replicative errors as they crop up, and to cause cells that cannot be repaired to commit suicide (apoptosis) before they can become cancerous. When our inherited germline DNA repair genes are faulty, cancers may appear at any time and grow unchecked. They also can result in radio-toxicity because healthy cells can't fix the X-ray damage to their DNA and die off en masse. This is the case for ATM and ATR mutations that occur in both copies of the inherited genes (called "bi-allelic") When tumor DNA repair is faulty, as it often is, the cells become immortal, DNA errors proliferate and lead to such phenomena as EMT (cells able to exist outside of the prostate and migrate easily), castration resistance, and drug resistance.

The table below shows the incidence of several of the most important DNA-repair genes and their prevalence (1) in men with metastatic prostate cancer (2) in men with localized prostate cancer, and (3) in men in the general population who don't have prostate cancer. About 1 in 8 (12%) men who have detected metastases have a germline DNA repair defect. That falls to only about 1 in 22 men who have localized prostate cancer, and 1 in 37 men without prostate cancer.

It is only worth knowing about if there is something we can do about it. Someday we may have gene editing tools that can correct those aberrations throughout the body. CRISPR and Zinc Finger technology are in their infancy, and have only just started to be used in clinical studies for prostate cancer. The two medicines we have in our armamentarium against prostate cancer in those with germline DNA-repair defects are PARP inhibitors (e.g., olaparib, rucaparib, talazoparib, etc.) and platinum-based chemotherapy (e.g., carboplatin, oxaliplatin, etc.). (Update Oct. 2019) Recent trials suggest that only those with the BRCA1/2 mutations (and maybe CHK2) respond to current PARP inhibitors see this link).

Color Genomics

Color Genomics is a division of Genome Dx, the same company that offers the Decipher test. They are now offering a 30-gene panel listing the most frequently observed mutations in DNA-repair genes. It includes all of the genes listed in the table above plus other genes that have been implicated in other cancers (see the list here). They also look for aberrant TP53 and PTEN - two gene mutations that have been implicated in the loss of tumor suppression and loss of apoptosis, and are prognostic for aggressive prostate cancer variants. What is astounding is the price -- only $249! A full genomic analysis of BRCA2 would cost somewhere between $2,000 to $3,000. By limiting  their analysis to the most common site mutations, they are able to make it affordable, albeit not as thorough. It can be ordered by a physician (they will provide one if necessary). It is a simple saliva test that the patient mails in, and genetic counseling is included with the results.

Associated with other indicators of poor prognosis

A team at Johns Hopkins reported on their use of the Color Genomics test in 150 patients to determine whether germline DNA-repair defects were associated with two rare and aggressive prostate cancer variants: ductal and intraductal prostate cancer. They also looked for associations with lymphovascular invasion discovered at pathology. Velho et al. reported:
  • Ductal/intraductal histology was discovered in 48% who had the defects vs only 12% who were free of those defects.
  • Lymphovascular invasion was discovered in 52% who had the defects vs. only 14% who were free of those defects.
  • 44% of patients with a positive germline test would not have been offered genetic screening according to current National Comprehensive Cancer Network (NCCN) guidelines. (update note: NCCN has changed its guidelines)
Other tests

While 23andMe offers a germline test that the consumer can order without a doctor, it is inferior. There are, say, 10,000 or more genetic mutations that can occur within a single BRCA2 gene. 23andMe only looks at a narrow pre-defined range of genomic abnormalities, using a silicon SNP array. Color Genomics uses "next generation sequencing" to look at many more types of genomic aberrations. There are other tests available from AmbryGenetics and Myriad.

Those who test positive may wish to investigate a clinical trial of a PARP inhibitor:

Carboplatin trials specifically for men with DNA-repair defects:

Most (7 out of 8) metastatic patients will learn nothing from this test and it will be a waste of money. But for some who seem to have an unusually aggressive prostate cancer variant, have ductal/intraductal histology, or have had lymphovascular invasion identified at pathology, it may be worth paying for the relatively inexpensive test. It may indicate that a platin may be a preferred form of chemo, or that a clinical trial of a PARP inhibitor may be warranted.

Saturday, November 18, 2017

Myth: "Gleason 6 never progresses"

There is a lot of mythology about prostate cancer. One of the prevalent myths is that a Gleason score 6 (GS6) found with a biopsy, or even confirmed over many biopsies, never progresses. How did the myth get started and what is the truth, as we know it so far?

Metastases never come directly from GS6 in the prostate (true)

This is true. Donin et al. at NYU Langone looked at the records of 857 patients diagnosed as GS6 after a prostatectomy. 16 of them were found to have a significant recurrence and were treated with salvage radiation to the prostate bed. All but 2 of the treated patients had no further recurrence, indicating that there were no distant metastases. The remaining two were found to have actually had GS7 when their removed prostates were re-examined.

Ross et al. looked at records from Johns Hopkins, UCSF,  Baylor and Henry Ford and found 22 cases (out of 14,123 examined) where pelvic lymph node metastases were found during prostatectomies of GS6 patients. 19 of those were re-reviewed, and all were found to have a higher Gleason score than the initial pathology assessment. Lymph node metastases were never associated with a GS6. Similarly, Liu et al. searched the 2004-2010 SEER database to find 21,960 GS6 patients who had pelvic lymph node dissection along with their prostatectomy. Only 0.48% were found to have lymph node metastases. Unfortunately, their prostate specimens were not available for re-review. Wenger et al. looked at the 2004-2011 SEER database, the 2004-2013 National Cancer Database, and  2004-2013 patient records at the University of Chicago; lymph node metastases were found in 0.2%, 0.18%, and 0%, respectively, among the GS6, post-prostatectomy records. Of the 24 patients at U. of Chicago who had a recurrence, all but 3 were local. The 3 non-local recurrences were all found to have been GS7 on pathological re-review.

In a Dutch study of 449 GS6 post-prostatectomy patients treated from 1985-2013 with over 8 years of median follow-up, Kweldam et al. found that there were no lymph node metastases, no distant metastases, and no prostate-cancer related deaths.

Not only does true GS6 never metastasize, it rarely eats into surrounding tissue. Anderson et al. looked at post-prostatectomy records of 2,502 GS6 patients from the University of Chicago and Northwestern treated from 2003-2014. Only 7 of them were found to exhibit extraprostatic extension (stage pT3a), and it was only focal in every case. There were no cases of seminal vesicle invasion (stage pT3b) or invasion of organs adjacent to the prostate (stage pT4). Hassan et al. at Johns Hopkins looked at post-prostatectomy records of 3,288 GS6 patients treated from 2005-2016. 3.9% exhibited focal extraprostatic extension, 2.4% exhibited significant (non-focal) extraprostatic extension, and there was only one case of seminal vesicle invasion.

A GS6 at biopsy may not be a true GS6 (true)

In a study at Johns Hopkins among low-risk patients who decided to have a prostatectomy, Epstein et al. reported that 36% were upgraded from a GS6 to a higher grade. Lotan et al. report a similar amount of upgrading (40%) even if a 32-core saturation biopsy was used. Multiparametric MRI targeting can find GS7 or greater tumors if they are large enough (> 2 cc), but Bratan et al. reported that the GS7 detection rate was only 63% for tumors < 0.5 cc, and 82-88% for tumors 0.5-2.0 cc. An NIH study used mpMRI imaging on patients who immediately afterwards had a prostatectomy. The mpMRI was evaluated by their expert readers who looked for cancers larger than 5 mm and  with grades of at least GS 3+4. The mpMRI missed 16% of clinically important tumors.

Multiple biopsies over the years can increase the odds of finding any cancer that is GS7. Even if a single mpMRI-targeted biopsy misses 37% of small GS7 tumors, the odds that two such biopsies will miss it is 37% x 37% = 14%. Three biopsies drop the odds to 1 in 20, and the odds of missing it on 4 biopsies is 1 in 53. Multiple biopsies are used in most active surveillance protocols, and their frequency can be slowed if there is no evident progression. Some of the increased detection with multiple biopsies will be due to better detection of what was always there, some will be due to grade progression (see below), and it really doesn't matter which is which. 

A GS6 may progress into something else that can metastasize (true)

GS6 is a relatively indolent type of prostate cancer. Most of it never progresses to higher grades, but some of it will, given enough time. In the longest-running active surveillance study in North America, Klotz reports that 55% of low-risk patients have been able to avoid treatment due to grade progression for 20 years. Conversely, 45% did have grade progression. Only 25% had progressed in the first 5 years, 37% by 10 years, and 45% by 15 years. After that, cases of progression came to a halt. An  active surveillance model at Johns Hopkins, which had strict annual biopsies, attempted to separate the misclassifications from the true grade progression. They estimated that the true grade progression rate in the first 10 years was 12%-24%. A similar model estimated the rate of total grade progression (true progression plus correction of prior misclassification) at about 4% per year during the first ten years, and they determined that the time for those GS6s that progressed to a GS7 took an average of 14 months.

Watkins et al. reported that GS6 patients had a low risk of recurrence after prostatectomy unless they had positive surgical margins. 8-year freedom from biochemical recurrence was 95% with no positive margins, but only 74% if there were any positive margins. GS6 left in the body can still proliferate and progress.

A large retrospective study at Harvard reported that among the 410 deaths from prostate cancer in an advanced prostate cancer cohort, 42 (10%) were originally biopsy-diagnosed as GS 6. GS 6 doesn't often turn into something lethal, but it can.

Further evidence of grade progression comes from a cohort of 1041 Swedish patients who had a PSA test but were not biopsied at that time. Gleason score was found to be correlated with the "lead time" between the date of the elevated PSA (3.0-10.0) and the biopsy when symptoms occurred. The diagnosis of higher grade cancers rose steadily throughout the up to 30 years of lead time in the study. Inversely, the diagnosis of low grade cancers dropped steadily with lead time. The low grade cancers converted to high grade cancers over time.

(update July 2020) Salami et al. found that certain molecular "fingerprints" existed in the cancers that were GS6 and remained in their cancers after the patients progressed to higher grades.

There are some risk factors that can help distinguish the GS6s that will progress from the ones that won't (partly true)

There are several risk factors associated with higher grade cancers (e.g., PSA, PSA density, age, and African-American); however, none of them have a cutpoint that discriminates between those who will progress from GS6 and those who won't. Ellis et al. at Johns Hopkins showed that the number of positive biopsy cores (≤6 or >6) did predict grade progression at prostatectomy to some extent:
  • 23% were upgraded if there were ≤6 positive cores
  • 34% were upgraded if there were >6 positive cores
  • But it had little effect on the recurrence-free survival following prostatectomy . 
Perineural invasion (PNI) noticed in the biopsy may be prognostic for progression on GS6, especially when both PNI and a high percentage of cancer in cores are found (see this link). Johns Hopkins has tables from which a patient with PNI may estimate his risk that the cancer is not confined to the prostate capsule based on GS, PSA, and the highest % cancer in a biopsy core.

Genetic risk can sometimes identify patients who are at higher risk for grade progression than their low risk NCCN designation would indicate. Prolaris and Oncotype Dx will usually indicate genetic risk in line with NCCN risk group, but sometimes they may find higher risk than expected. Recently, Decipher has begun to offer genomic risk scores based on biopsy samples. There is some evidence that there may be wide genetic diversity of the multiple tumors within a man's prostate. Just as a single biopsy may miss a high grade cancer, the biopsy cores sent for genetic analysis may not include the one or ones with higher genetic risk. These tests are expensive ($3,000-$4,000) and may not be covered by insurance (always get preauthorization!).

PSA inconsistently goes up with grade, and some high grade disease puts out low levels of PSA. Compared to total PSA, the PSA-derived biomarkers (e.g., Free PSA, PHI, 4Kscore, IsoPSA) have higher detection rates of high grade prostate cancer, as do PCA3, TMPRSS2:ERG fusion, and Select MDx.

How does GS6 progress to higher grades? What can we do about it?

There is very little certainty about the changes that occur at the molecular level when GS6 cancer progresses, and what drives those changes. Sowalsky et al. found that Gleason pattern 4 glands that were intermixed or adjacent to Gleason pattern 3 glands shared characteristic genetic markers that indicated they had a common origin. Whether the pattern 4 cells were cloned directly from pattern 3 cells or arose from  a common precursor cell is not clear. The authors suggest that a genetic aberration called "PTEN loss" may distinguish pattern 3 cells that might progress from the kind that might not progress.

VanderWeele et al. did a genetic analysis of 4 patients who had both GS6 and GS8 tumors in the same prostate. Two of the men also had lymph node metastases that were analyzed. They found that the GS6 and GS8 cells shared 9% of the characteristic genes they looked for. That suggested that GS6 did not directly morph into GS8, but arose from an early common ancestor long before. On the other  hand, the GS8 shared 81% of those characteristic genes with the lymph node metastases, indicating recent progression.

Haffner et al. did a genetic analysis of a metastasis from a patient who recently died of prostate cancer. They also analyzed tissue samples taken from the man's earlier prostatectomy and lymph node dissection. They found that the lethal metastasis was much more closely related to a small bit of pattern 3 cancer in the prostate rather than to the pattern 4 cancer in the "index lesion" (the largest, highest grade tumor in the prostate) or to the lymph node metastasis. The lymph node metastasis, however, was not clonally related to the pattern 3 cancer, and seems to have arisen from a different source. The authors suggest that PTEN loss and loss of another tumor-suppressor gene called TP53 may distinguish the potentially lethal pattern 3 cancer from the innocuous kind.

Palapattu et al. used MRI/Ultrasound fusion biopsy to take sequential cores from exactly the same place on study entry and one year later from 31 low-risk men on active surveillance. 35% progressed to a higher grade (pattern 4 or 5) in the same site in that year. It was from the same tumor because it had characteristic gene expression in almost every case. They found several suspicious genetic mutations. Mutations in the genes SPOP and IDH1 were common to both the low grade and high grade cells in one patient each, suggesting they may be responsible for progression. In one patient, a TP53 mutation was found in the later high grade core, but not the earlier low grade core, suggesting it is the result of something else that caused progression. Mutations in SPOP and BRCA2 were found in only the later cores in two patients who did not yet progress, perhaps suggesting increased potential for progression. As opposed to the studies that suggest a common progenitor cell for low and high grade cancer, this study suggests that genetic abnormalities and accumulating genetic breakdown are the sources of grade progression.

These studies are beginning to offer clues as to why GS6 sometimes progresses. There are biomarkers like Ki67, p53, and VPAC that may predict progression. Histological analysis may detect PTEN loss and TP53 loss, as well as mutations in SPOP, IDH1, and BRCA2 in tumor genes. While we currently lack widely available means to predict progression, active surveillance with periodic biopsies remains our best tool for finding progression while it is still curable.

Unfortunately, there are no medicines that can prevent grade progression from occurring, or reverse it after it does occur. Perhaps someday, CRISPR or zinc fingers will be used, but that is many years away.

Tuesday, November 1, 2016

PORTOS: a gene signature that predicts salvage radiation success

Salvage radiation is curative in roughly half of all cases. There are many factors that contribute to an unfavorable prognosis, including waiting too long, high PSA and rapid PSA doubling time, adverse post-surgery pathology (stage, Gleason score, positive margins), and high Decipher or CAPRA-S score. But, other than a detected distant metastasis, none can predict failure of salvage therapy. For the first time, there seems to be a genetic signature that predicts when adjuvant or salvage radiation  (A/SRT) will succeed.

The study is all the more impressive because of the many top prostate cancer researchers attached to it, representing a collaborative effort from many top institutions: Harvard, University of Michigan, Johns Hopkins, Northwestern University, University of California San Francisco, Mayo Clinic and others.

The process

Zhao et al. started with data on 545 patients who had a prostatectomy at the Mayo Clinic between 1987 and 2001. They attempted to find patients who were matched on pre-RP PSA, Gleason score, stage, and positive margins, but differed on whether they received A/SRT or not. They also had to have complete information on diagnosis and whether they eventually had metastatic progression. This yielded 98 matched pairs. They then did complex genetic screening of archived tissue samples from those prostatectomy patients, focusing on 1800 genes that have been implicated in response to DNA damage after radiation. They found 24 genes that were correlated with occurrence of metastases after salvage radiation. After correcting for other factors, they determined what they call a “Post Operative Radiation Therapy Outcomes Score (PORTOS).” A PORTOS of zero (called a “low” PORTOS) means it predicts no benefit from salvage radiotherapy. A PORTOS greater than zero (called a “high” PORTOS) predicts a benefit from salvage radiation.


The next phase was to predict how well the 24-gene signature would predict salvage radiation success in a larger data set. They analyzed 840 patient records from patients treated at the Mayo Clinic from 2000-2006, Johns Hopkins (1992-2010), Thomas Jefferson University (1999-2009) and Durham VA Medical Center (1991-2010). They were able to find 165 matched pairs – half treated with A/SRT, half with no radiation. Tissue samples were screened and scored, and 10-year incidence of detected metastases was obtained. 1 in 4 men were categorized as “high PORTOS,” 3 in 4 were “low PORTOS.”

In the “high PORTOS” group: 
  • Only 4% suffered metastatic progression if they had A/SRT
  • 35% suffered metastatic progression if they did not have A/SRT
  • They had an 85% reduction in 10-year incidence of metastases after A/SRT, which was statistically significant.
In the “low PORTOS” group:
  • 32% suffered metastatic progression if they had A/SRT
  • 32% suffered metastatic progression if they did not have A/SRT
None of the other prognostic tools (Decipher, CAPRA-S, or Prolaris) that are sometimes used to predict metastases after prostatectomy could predict the response to A/SRT.


This should be interpreted with caution for several reasons:

It was retrospective, and therefore subject to selection bias. That is, the physicians may have decided on the basis of patient characteristics or other disease characteristics not captured here to give A/SRT to some patients, but not to others. Only a prospective, randomized trial can tell us if the association with PORTOS is the cause of the differential response.

Among the disease characteristics the researchers were unable to capture for this study were the time between prostatectomy and A/SRT, PSA at time of A/SRT/maximum PSA reached, nadir PSA achieved after prostatectomy, PSA doubling time, extent of positive margins, Gleason score at the positive margin, and comorbidities. Patients were not treated uniformly with respect to radiation dose received and duration of adjuvant androgen deprivation therapy (ADT). Only 12% received any adjuvant ADT, and only 12% received adjuvant (rather than salvage) radiation.

Metastases were detected by bone scan and CT. Lymph node dissection, if performed, was limited. It was detected in 4% of the “low PORTOS” group, but in none of the “high PORTOS” group. It is unclear how today’s newer PET scans would affect outcomes.


Prostate cancer has long been known to be radioresistant relative to other cancers. To understand radioresistance, we must first understand how ionizing radiation (X-rays or protons) kills cancer cells. The radiation causes a chemical reaction with water and oxygen to generate molecules known as “reactive oxygen species” or ROS. One such ROS molecule, the hydroxyl radical, inserts itself into the cell’s DNA to break both strands of the double helix, called “double strand breaks.” The cell dies when it can’t replicate because of those double strand breaks.

Radiobiologists cite 5 reasons for radioresistance:

1. Hypoxia

Prostate cancer thrives in an oxygen-poor environment, and often does not have a good blood supply that brings oxygenation. It therefore requires more radiation to provide adequate ROS, especially into thick tumors.

2. Cell-Cycle Phase

As a cancer cell attempts to build new DNA and replicate, it goes through several phases. In one of those phases, the “S phase,” the cell is building new DNA. It is particularly radioresistant in this phase. Radiotherapy is typically carried out over a period of time in multiple fractions, rather than in a single shot, to allow the cancer cells to cycle into more radiosensitive phases. However, in a recent lab study, McDermott et al. showed that fractionated radiation increases the population of radioresistant S-phase prostate cancer cells.

3. Repair of DNA damage

Non-cancerous cells that can’t repair the DNA damage, commit suicide (called apoptosis). Many non-cancerous cells are able to repair the DNA damage and survive. Fractionation gives them time to self-repair. Cancerous cells usually lack that DNA-repair mechanism and most cannot undergo apoptosis. If they are not killed immediately, they die when they try to replicate. However, some cancerous cells may escape destruction by turning the genetic cell repair mechanism back on.

4. Repopulation

Some cancers grow so quickly that fractionated radiation gives them time to grow back between treatments. This is not the case for prostate cancer.

5. Inherent radioresistance

Some kinds of cells are inherently impervious to radiation damage; muscle, nerves, and stem cells are radioresistant, as are melanoma and sarcoma. Prostate cancer stem cells, thought to play a role in prostate cancer proliferation, are inherently radioresistant. A recent lab study showed that radiation may paradoxically activate stem-cell like features of prostate cancer cells, turning them into radioresistant stem cells.

How should PORTOS be used?

GenomeDx is already supplying PORTOS to post-prostatectomy patients who order Decipher. Should it be used to guide A/SRT decision-making? Given the caveats (above), there are many uncertainties in how predictive it actually will be when it is used prospectively in larger patient populations. But the information is certainly interesting.

I wonder whether PORTOS reflects a genetic change that occurs in local prostatic cancer cells as they undergo a change (called “epithelial-to-mesenchymal transition” (EMT)) into metastatic-capable cells. Or is it a genetic characteristic, there from the start? A recent study showed that 12% of men with metastases have faulty DNA-repair genes. (This included 16 DNA-repair genes, compared to the 24 in the PORTOS study). Such faults occurred in 5% of men with localized prostate cancer, and 3% in men with no prostate cancer. DNA-repair mutations seem to accumulate as the cancer progresses. It may well be that PORTOS is an early detector of systemic micrometastases. Perhaps it will be found to be redundant to detection of small metastases using new PET indicators. I would love to see a PORTOS analysis on metastatic tissue as well (lymph node, bone and visceral) and maybe on circulating tumor cells to see whether radioresistance is an acquired trait of PC progression. If it is an early indicator of metastatic progression, it may already be too late for primary radical therapy.

While a “high” PORTOS suggests that A/SRT will be curative, only a quarter of the men had a high PORTOS. Does that really mean that three-quarters of recurrent men should give up on curative therapy? If PORTOS is not an indicator of EMT, I hope that those recurrent cancers still can be cured. But it may mean that certain adjuvant measures may be required, including higher radiation doses, systemic therapies that are known to enhance radiation effectiveness, and investigational adjuvant therapies.

      A/SRT doses are typically in the range of 66-70 Gy. Some A/SRT studies used doses as high as 72-76 Gy. With modern IGRT/IMRT technology, such doses may be delivered with acceptable toxicity. Also, if larger lesions can be identified with the new PET scans and multiparametric MRIs, it may be possible to deliver a simultaneous integrated boost dose to those lesions.

      ADT has been shown to reduce hypoxic cancer survival and inhibit DNA repair. It is possible that prolonged neoadjuvant use, perhaps with second-line hormonal agents (Zytiga or Xtandi) may improve radiation cell kill. Docetaxel, which has shown limited usefulness in non-metastatic patients, may prove useful in low-PORTOS situations. Perhaps immunotherapy can play a role as well.

    There are many investigational agents that may enhance radiosensitization. PARP1 inhibitors (e.g., olaparib) and heat shock protein inhibitors may prove useful in restoring radiation sensitivity (see this link). PI3K/mTor inhibitors and HDAC inhibitors (e.g., vorinostat) may increase cell kill in hypoxic conditions (see this link) and to cancer stem cells (see this link). Cell oxygenation may be enhanced by a measure as simple as 15 minutes of aerobic exercise before each treatment (see this link). There are common supplements like resveratrol and soy isoflavones, and drugs like statins, aspirin, and metformin that have shown promise as radiosensitizers in lab studies.

It is possible that PORTOS may also prove useful in predicting radiation response among newly diagnosed unfavorable risk patients. GenomeDx  currently requires whole-mount prostate specimens. I don’t know if PORTOS can be done on biopsy cores, or if it provides any prognostic information beyond what the conventional risk factors (PSA, Gleason score, stage and tumor volume) provide. It would have to be similarly validated before we would be able to incorporate it in primary therapy decision-making.

This test is very expensive. For now it only is available along with Decipher, which costs about $4,000. Medicare may cover it, but private insurance may or may not. Always get pre-authorization first.